Multiservice provisioning has been a development trend of the field. At present, there are two types of architecture which support multiservice, single-edge architecture, and multi-edge architecture.
Support for multiservice scenarios by conventional single-edge and multi-edge techniques will be described hereinafter with reference to FIG. 1 and FIG. 2, respectively.
FIG. 1 illustrates a structural diagram of a conventional, in the case of single-edge, supporting multiservice. As illustrated in FIG. 1, the conventional single-edge technique supports a multiservice scenario.
As can be seen from FIG. 1, an access node (AN) 104 corresponds with a single broadband remote access server (BRAS) 102 to which all service providers, i.e. service providing nodes, are connected. The access server controls user selections of service providing nodes 106 and processes subsequent service flows. Once a new service is added, a corresponding feature support should be added at the access server 102. Authentication of a user 108 and control of selections of service providers 106 are also done at the access server 102.
The single-edge technique illustrated in FIG. 1 has the following disadvantages: because of the variety of service features provided by different service providers, the access server is required to support every service feature, and control flows, such as authentication and accounting, of all users will pass the access server; therefore, the access server is required to support numerous functions, which leads to poor extensibility, and becomes a bottleneck of the whole network.
FIG. 2 illustrates a structural diagram of a conventional, in the case of multi-edge, supporting multiservice. As illustrated in FIG. 2, the conventional multi-edge technique supports a multiservice scenario.
As can be seen from FIG. 2, broadband network gateways (BNGs) 202 are edges of the access network. Selections of service providers, i.e. service providing nodes 206, are done by an AN 204, and related functions such as authentication, authorization, accounting, policy distribution, and Internet Protocol (IP) address allocation are supported by the BNGs 202. The benefit of multi-edge technique is that, different BNGs can be provided to implement different types of services, which makes services easy to be extended.
The multi-edge technique illustrated in FIG. 2 has the drawback that, the BNGs not only forward services, but also perform authentication and control of services. In the case of multi-edge, these control functions are separated among each BNG, so that centralized control of the access network is difficult to achieve. In addition, the AN would be difficult to implement because it is required to have the AN implement the function of network selection.
In a single-edge architecture, a BRAS is the network edge node at which user authentication, authorization, and control are performed collectively. The BRAS has a single connection with an AN, and can perform QoS control of the AN based on a policy. The BRAS also connects multiple service providing nodes, selections of the service providing nodes, and support for various services are all implemented on the BRAS. As the only edge control node, the BRAS is also the only node where various edge services are initiated. Accordingly, the network edge node is the only device in the access network that implements both control and bearing functions; and that the network edge node is required to support a variety of services. Therefore, in the case of single-edge, the functions of the network edge node are complex, difficult to be implemented or extended, and easy to cause single point of failure.
However in a multi-edge architecture, different network edge nodes correspond to and can be optimized for different services. Such a multi-edge architecture is good for extensions of services, and simplifies the implementation of network edge nodes. But new problems of centralized control of users by network edges and selections of network edge nodes by users are raised. Because of the variety of network edges, it would be a problem for the edge nodes to coordinate user control; and that it is required by the architecture for an AN to select network edge nodes, which increases the complexity of the implementation of the AN, meanwhile the implementation of control functions by the edge nodes is not simplified.